Background
Embryonal carcinoma (EC) cells are the undifferentiated and pluripotent component of germ cell nonseminoma tumors. Some EC cell lines can be induced to differentiate in response to cellular or pharmacological morphogens. These cells share many features in common with embryonic stem (ES) cells and their induced differentiation mimics critical stages of early embryogenesis [
1]. Additional evidence indicating that EC and ES cells are closely related comes from their shared gene expression profiles, which are highly specific to germ cells and pluripotent ES cells [
2]. These species include the transcription factors POU5F1 and Nanog, bone morphogenetic protein family member GDF-3, developmental pluripotency-associated gene 3 (DPPA3) and fibroblast growth factor 4 (FGF4).
The Wnt signalling pathway is essential for normal eukaryotic development and inappropriate activation of Wnt signalling occurs in many cancers [
3]. Wnt ligands engage signal transduction through multiple receptors including the Frizzled transmembrane receptor family, co-receptors LRP5 and LRP6 and receptor tyrosine kinases, Ryk and ROR2 [
4]. There are 19 Wnt ligand and 10 Frizzled receptor genes in the mammalian genome. The canonical Wnt-Frizzled signalling pathway results in stabilization of β-catenin allowing it to enter the nucleus and activate transcription of Wnt target genes by binding to T-cell factor/lymphoid enhancer factor (TCF/LEF) [
5]. Frizzled receptors also play a key role in the planar cell polarity (PCP) pathway that is responsible for orienting cells relative to each other, and in a G protein-dependent pathway that triggers the release of calcium (Ca
2+) [
5]. The other Wnt receptors Ryk and Ror2 can signal through Src and JNK intermediates, respectively [
6]. Wnt signalling proteins promote expansion of stem cells in diverse tissue contexts including the mammary gland, hematopoietic system, and the brain, underscoring the importance of this signalling pathway in stem cell maintenance [
7].
The multipotent EC cell line NT2/D1 differentiates along a neuronal lineage in response to all-
trans retinoic acid (RA) treatment, which is associated with loss of both self-renewal capacity and expression of pluripotent specific genes [
8]. NT2/D1 cells were derived from a metastasis of a human testicular germ cell tumor (TGCT) and these retain the pathognomonic cytogenetic marker and cellular features of this malignancy [
1,
9]. In our initial studies to identify key species regulating early differentiation steps, several components of the Wnt signalling pathway were affected by RA-treatment [
8]. This study sought to build on that prior work by comprehensively examining the expression and activity of Wnt species during induced differentiation of NT2/D1 cells and in a well characterized panel of TGCT cell lines including a derived RA-resistant cell line, NT2/D1-R1 [
10]. Given that this pathway is important for both the maintenance of pluripotency and in regulating specific differentiation steps, it was hypothesized that during induced differentiation, the Wnt signalling machinery was reprogrammed in EC from a pathway supporting pluripotency to one promoting differentiation. Findings reported here provide substantial evidence confirming this hypothesis and these implicate the therapeutic potential of targeting the Wnt pathway in human EC and other TGCTs.
Methods
Cell Culture and Clonal Growth Assays
NT2/D1, NT2/D1-R1, Tera-1, 833K and 293T human cells were cultured in high glucose DMEM (InVitrogen, CA) containing 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM glutamine, and 10% fetal bovine serum (FBS) at 37°C under humidified 5% CO2. RA was dissolved in dimethyl sulfoxide (DMSO) as a 10 mM stock solution and stored in liquid nitrogen. Cells were treated with RA (10 μM), which was refreshed every 48 hours, while control cells were treated with DMSO vehicle alone. For colony formation assays, NT2/D1 and NT2/D1-R1 were independently plated at 5000 and 2500 cells per well, respectively, in six well tissue culture plates in triplicate. After 10-14 days, colonies were stained using Diff Quick (Imeb, Inc., CA) and quantified using the ColCount instrument (Oxford Optitronix, UK). Replicate experiments were performed to confirm results.
siRNA Silencing
Double-stranded small interfering RNAs (siRNAs) with 19-nucleotide duplex RNA and a 2-nucleotide deoxythymidine overhang at the 3' region were synthesized (Dharmacon, CO and Qiagen, CA). Two different siRNAs were designed to target independently human POU5F1, FZD5 and FZD7 mRNAs. The sequences chosen were: FZD5-1 5'-ATCACGGTGCCCATGTGCC-3', FZD5-2 5'-TCCTAAGGTTGGCGTTGTA-3', FZD7-1 5'-AACGGCCTGATGTACTTTA-3', FZD7-2-2 5'-GGCCAGCTTGTGCCTAATA-3', POU5F1-2 5'-AGCAGCTTGGGCTCGAGAA-3' and POU5F1-3 5'-GAAAGAACTCGAGCAATTT-3'. Control siRNAs used were firefly luciferase pGL2 siRNA (5'-CGTACGCGGAATACTTCGA-3') and RISC free (5'-GCGCGCTTTGTAGGATTCG-3'), the latter is not a substrate for the RNAi silencing complex (RISC) (Dharmacon, CO). Transfection of siRNAs was accomplished with OligofectAMINE (Invitrogen, CA), as previously described [
11]. To assess effects on cell growth, EC cells were trypsinized 24 hours after transfection and plated at 5 × 10
4 cells per well in 6-well plates. Triplicate wells were counted for each time point post-transfection and each experiment was performed at least three independent times. Results were expressed as mean ± standard deviation for a representative experiment. Subsequent assays for growth of small hairpin RNA (shRNA)-transduced cells were plated in the same way, but cell growth measurements were performed using the Cell Titre-Glo Assay (Promega, WI), which uses ATP content as an indicator of cell number.
Immunoblot Analysis
Cells were lyzed in a modified radioimmune precipitation buffer, as described [
12]. Protein concentrations were determined using the BCA protein assay kit (Thermo Scientific, IL). Proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE) and transferred to nitrocellulose membranes. Antibodies used and dilutions were FZD5, 1:500 (06-756 Upstate, NY), FZD7, 1:250 (AF198 R&D Systems, MN), active-β-catenin (ABC) 1:500 (05-665, Upstate, NY), β-catenin 1:2000 (MD, 610153, BD Transduction Laboratories), JNK and phospho-JNK 1:1000 (9258 and 9251, Cell Signaling Technology, Inc. MA), Oct3/4/POU5F1, 1:750 (H-134, Santa Cruz, CA), α-tubulin 1:1000 (CP06, Calbiochem, CA) and actin 1:2000 (C11, Santa Cruz, CA). Membranes were blocked in tris-buffered saline with 0.1% Tween 20 (TBS/Tween) plus 5% non-fat milk powder and primary antibodies were also diluted in 5% milk powder in TBS/Tween except for the ABC antibody which was diluted in 1% milk powder as previously described [
13] and the JNK antibodies that were in 5% bovine serum albumin. Primary antibodies were detected with horseradish peroxidase conjugated secondary antibodies (Santa Cruz, CA and Amersham, IL) and visualized using the enhanced chemiluminescence system (Amersham, IL). For FZD5 and FZD7 detection, cellular lysates were not frozen and run under non-reducing conditions (without β-mercaptoethanol or boiling).
Lentivirus Production and Luciferase Assays
The Topflash lentiviral vector was obtained from Dr. Karl Willert (University of California, San Diego). Silencing shRNA constructs were purchased (Open Biosystems, AL). To generate lentiviral particles, the desired plasmid and lentiviral packaging plasmids pCMV-dR8.2 dvpr and pCMV-VSVG (Addgene.com) were transfected into 293T cells using FuGENE 6 (Roche, Germany). Medium was changed 18 hours after transfection and viral particles were harvested twice over the next 72 hours. Cells were infected with the lentiviral supernatant for 24 hours in the presence of 4 μg/ml of polybrene (Sigma-Aldrich, MO). After 48 hours, cells were replated in 10-cm dishes. For shRNA lentiviral constructs, infected cells were selected by addition of 0.5 μg/ml puromycin (Sigma-Aldrich, MO). The Topflash lentiviral vector does not carry an antibiotic selectable marker and pools of cells were maintained for subsequent analyses. Luciferase activity was measured using the Promega (WI) luciferase assay system and normalized to protein concentrations, as above.
Analysis of Wnt Signalling Pathway Genes
To assess regulated expression of Wnt pathway members following RA-mediated differentiation of NT2/D1 cells, real-time quantitative reverse transcription-polymerase chain reaction (RT-PCR) assays were performed. The RT2 Profiler™ PCR Array PAHS-043 (SABiosciences, MD) was used for these analyses. Total cellular RNA was isolated using TriReagent (Molecular Research Center, OH) and cDNA was synthesized from 1 μg of total RNA according to the manufacturer's instructions (SABioscience). Thermal cycling was performed using an Applied Biosystems 7500 Fast Real Time PCR System and real-time amplification data were obtained using ABI 7500 Fast System SDS software. Gene expression was normalized to internal controls (housekeeping genes) to determine fold changes in gene expression between test and control samples (SABioscience, MD). For the profiler arrays, the SYBR green PCR mix used was from -SABiosciences (MD). For real-time RT-PCR analyses of individual gene products, cDNA synthesis and SYBR green PCR reagents were purchased from Applied Biosystems (CA). Primer sequences used for real-time RT-PCR assays appear in additional file
1.
Differentiation Marker Studies
Indirect fluorescence-activated cell sorter (FACS) analyses to evaluate RA-induced neuronal differentiation of NT2/D1 cells were performed using previously established methods [
14]. Briefly, NT2/D1 cells were harvested by trypsinization and incubated with monoclonal antibodies to established human EC differentiation markers. The A2B5 antibody was isolated from a hybridoma culture and recognizes a neuronal epitope in RA-treated NT2/D1 cells (ATCC, VA). SSEA-3 (Developmental Studies Hybridoma Bank, University of Iowa, U.S.A.) recognizes an embryonal antigen, which is lost as EC cells differentiate [
15]. Cells were indirectly assayed with Alexa 488 nM conjugated goat anti-mouse or goat anti-rat antibodies (Invitrogen, CA) and fluorescence was measured, as described [
14].
Statistics
Where a value for statistical significance is indicated a two sample two-tailed t-test assuming unequal sample variances was performed.
Discussion
Canonical Wnt signalling is associated with proliferation and is deregulated in cancer [
3]. Wnt target genes include cyclin D1 and c-myc, both of which are also associated with proliferation and are over-expressed in several cancers [
3]. Human NT2/D1 EC cells have high levels of cyclin D1 and c-myc, which are both repressed as these cells are induced to differentiate [
16,
29]. Based on this, it was hypothesised that canonical Wnt signalling would be active in the proliferative and pluripotent EC cells and repressed in these cells induced to differentiate with a concurrent loss of tumorigenicity. In fact, the opposite was observed with canonical Wnt signalling at low levels in basally growing cells and elevated levels as cells were induced to differentiate, exemplifying the paradoxical nature of Wnt signalling in ES and EC cells. This is also reflected in prior work where some studies report that canonical Wnt signalling supports ES cell self-renewal and pluripotency and others have found that canonical Wnt signalling can induce ES cell differentiation, as reviewed [
30]. Different cell lines and methodologies were used in these reports and might account for the apparent discrepant results. Also, the level of nuclear β-catenin and the origin of ES cells might determine how these cells respond to canonical Wnt signalling activation.
The EC cells examined here respond in a similar way to murine embryos in which Wnt signalling is not evident until day 6.5 when reporter activity is seen at the posterior of the embryo where the primitive streak begins to form [
31]. Embryoid bodies generated from Wnt reporter carrying ES cells also showed polarized Wnt activation, and the Wnt responsive cells were enriched for mesoendodermal gene expression [
31,
32]. Using a similar functional reporter assay in ES cells, it was found that transcriptional activity of the canonical Wnt pathway is minimal until the cells are induced to differentiate [
33]. In immunohistochemical studies of pediatric TGCTs, the EC specimens examined had low-levels of nuclear β-catenin while teratomas, which are the differentiated counterpart of these tumors, had elevated levels of nuclear β-catenin [
34,
35]. In combination with the observations reported here, this confirms the similarities in Wnt signalling between EC and ES cells.
From these analyses, several Wnt receptors and ligands were abundantly expressed in undifferentiated EC cells and repression of several of these species inhibited EC cell growth. WNT5A was the most abundant Wnt mRNA of the 16 examined in this study and this ligand has been most commonly associated with non-canonical Wnt signalling [
36]. It is likely that non-canonical Wnt pathways are functioning in human EC cells to maintain their cell division. The distinct chromatin structure of EC and ES cells may be responsible for the lack of β-catenin-mediated transcription and it has been proposed that one role of POU5F1 is to block canonical Wnt signalling [
17]. It was hypothesised that inhibition of WNT5A, the dominant Wnt ligand in the NT2/D1 system, would result in a more marked phenotype than repression of single frizzled receptors. As this was not the case, this indicates that either other Wnt ligands are playing redundant roles in this system and/or that the frizzled receptors are functioning without ligand, as reviewed [
4]. Additionally, while WNT5A mRNA levels were higher than that of other Wnt ligands, as assessed by real-time RT-PCR assays, it is not known if this relates to appreciably higher levels of Wnt protein or activity.
Many of the changes in Wnt gene expression seen in induced differentiation of EC cells are also seen in induced differentiation of ES cells, highlighting the use of the EC system as a model for Wnt signalling in ES cell biology. Other studies looking at POU5F1 target genes in human and mouse ES cells proposed that PORCN, FRAT2, and FZD5 were potential POU5F1 target genes [
37,
38]. In a bioinformatics study, POU5F1 transcription factor binding sites were conserved among mammalian FZD5 orthologs [
39]. Both SFRP1 and SFRP2 are downregulated in response to RA-treatment of EC cells [
8,
40] and in response to POU5F1 repression in murine ES cells [
37]. Of equal interest are the species strongly induced with both methods of differentiation, which include SFRP4, PITX2 and WNT2B. The induction of these species may be due to changes in chromatin resulting from the loss of POU5F1 activity or other mechanisms.
The dichotomous roles Wnt signalling plays in proliferation versus differentiation are especially evident in studies of neurogenesis. Differences in model systems and techniques may partially explain some of the discrepancies. However, it seems increasingly likely that the complexity of Wnt pathway allows for distinct signaling outputs within a cell. Recently, coactivator usage by β-catenin has been shown to determine proliferative versus differentiation gene transcription programs [
41‐
43] and manipulation of specific coactivator levels could maintain pluripotency in murine ES and EC cells [
42]. In addition to the increase in activated β-catenin seen as NT2/D1 cells are induced to differentiate, coactivator usage by β-catenin may also be changing and we aim to investigate this in future studies.
To target the Wnt signalling pathway, it was decided to specifically repress the expression of the FZD5 and FZD7 receptors. FZD5 was down-regulated in response to differentiation of NT2/D1 cells and in human melanoma cells FZD5 activation by WNT5A had been associated with elevated metastatic potential [
44]. FZD5 repression was growth suppressive in NT2/D1 cells, but was not as effective in the other TGCT cell lines tested. FZD7 was highlighted in a study to identify species preferentially expressed in ES and EC cells [
2] and was significantly growth suppressive in all the EC cell lines studied here. FZD7 has also been identified as a potential therapeutic target in hepatocellular and colorectal cancers [
45,
46].
Targeted repression of FZD7 was growth suppressive for all EC cells tested, but this did not cause differentiation of these cells. This is in contrast to a recent paper where shRNA-mediated repression of FZD7 decreased POU5F1 expression substantially in human ES cells and moderately in the human EC cell line NCCIT [
47]. These differences may be due to the distinct features of the cell lines used. Also, human ES cells may respond differently than human EC cells. A single shRNA construct was used in the human ES cell study, which caused a 60-fold knockdown of FZD7 mRNA, whereas the greatest knockdown achieved with our panel of shRNA constructs was 20-fold repression. In agreement between these reports were the observed phenotypic cell changes. For the human ES cells, this was particularly evident at the edges of colonies, whereas the NT2/D1 cells became elongated and spindle-like with repressed FZD7 expression (data not shown). Many of the non-canonical Wnt signalling pathways target the cytoskeleton, as in melanoma where non-canonical signalling by WNT5a is found to correlate with metastatic potential and invasiveness [
44]. Recently, it was reported that WNT5a polarizes the cytoskeleton, allowing chemokine directed motility in melanoma cell lines [
48]. TGCTs are highly metastatic and may be utilizing this pathway to enter the blood system and then spread to other sites. Subsequent studies aim to examine the role of Wnt pathways in EC cell motility and differentiation-induced cytoskeletal changes. Initial focus will be on DAAM1 and profilin-1, two proteins involved in actin organization and non-canonical Wnt signalling and up-regulated as EC cells are induced to differentiate along a neuronal lineage (Additional file
2 and [
49]).
The complexity of the Wnt signalling pathway is due to in part to the large number of potential ligands and receptors. In addition, challenges exist in developing pharmacological tools to specifically inhibit Wnt ligand activity that have made this pathway difficult to dissect. Wnt ligands were historically divided into those resulting in either canonical or non-canonical signalling. With the discovery of more Wnt receptors, non-Wnt ligands and Wnt-independent Frizzled activation, a new view of the Wnt pathway is being proposed: that it is the receptor and downstream effector context, in addition to ligand identity, which determines ultimate signalling output [
4]. The established role of Wnt signalling in development and its misregulation in cancer and other diseases highlights the need for a more complete understanding of how this pathway works. Novel pharmacological tools to inhibit Wnt signalling are becoming available [
50]. This will accelerate the evaluation of the therapeutic possibilities of Wnt pathway manipulation in human EC and other tumors.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
GES, ACK, AMB, KEE, ES, TB and SJF carried out molecular and cell biology experiments, and participated in the design of the study. GES and AMB prepared the figures and aided in interpretation of the data and in manuscript preparation. MJS provided the initial observations and helped in the design of experiments and in interpretation of results. ED contributed to the overall scientific direction, experimental design and interpretation, and in manuscript preparation. SJF directed the study and prepared the manuscript. All authors read and approved the final manuscript.